Download Infection elevates diversity - Aneil Agrawal

RESEARCH NEWS & VIEWS
Rubinsztein-Dunlop, H. Phys. Rev. Lett. 75, 826–829
(1995).
8. Uchida, M. & Tonomura, A. Nature 464, 737–739
(2010).
9. Harris, J. et al. Nature Phys. 11, 629–634
(2015).
10.Bazhenov, V. Y., Vasnetsov, M. & Soskin, M. S. JETP
Lett. 52, 429–431 (1990).
11.Bolduc, E., Bent, N., Santamato, E., Karimi, E. &
Boyd, R. W. Opt. Lett. 38, 3546–3549 (2013).
12.Verbeeck, J., Tian, H. & Schattschneider, P. Nature
467, 301–304 (2010).
E VOLUTIO NARY B IO LO GY
Infection elevates
diversity
Chromosomal shuffling in parental eggs or sperm can create new characteristics
in the next generation. In fruit flies, it seems that mothers with a parasitic
infection produce more such recombinant offspring than uninfected mothers.
A N E I L F. A G R A WA L
I
n most plants and animals, offspring are
genetically distinct from their parents.
Through the process of recombination,
which occurs as sperm and egg cells (gametes)
are produced, a parent can mix the two copies of a given chromosome received from its
own parents, thus transmitting unique chromosomes to its offspring. Writing in Science,
Singh et al.1 show that fruit flies produce a
higher frequency of offspring with recombinant chromosomes when the mother is
infected with a parasite than when it is uninfected. This intriguing observation may be an
important piece in the long-standing puzzle of
why recombination is so common.
Why should organisms shuffle their
genomes through sex and recombination? Natural selection should create an excess of good
gene combinations, so ‘undoing’ the work of
past selection by rearranging these genotypes
seems counterproductive. One possible explanation is that what constitutes a good combination of alleles (gene variants) changes over
time. In that case, undoing the work of past
selection is beneficial because selection in the
future demands something different. This idea
requires that selection on gene combinations
changes regularly2.
Coevolving natural enemies — particularly
parasites — might provide just the right type
of selection pressures for this scenario. This is
the basis for the ‘Red Queen’ hypothesis, which
proposes that sexual reproduction and recombination are favoured because they help hosts
to adapt to the ever-shifting selection imposed
on their gene combinations by the parasites3,4.
However, even rapidly evolving parasites do
not always induce selection for recombination; there are times in the coevolutionary
cycle when hosts are well adapted and nonrecombinant offspring will be more resistant
to infection than recombinant ones5,6. Intuitively, it might seem that the ideal solution is to
increase recombination when infected because
being infected indicates that your current gene
combination is not working.
To test this, Singh et al. performed hundreds
of test crosses using female Drosophila melanogaster fruit flies that carried mutations at
each of two genes on one chromosome, but
that had normal versions of the genes on their
other copy of the chromosome. The presence
of either mutation leads to visible physical
characteristics that allow determination of
whether one or both mutations are present in
their offspring — because the genes are in close
physical proximity, the normal or mutated
versions will be inherited together unless there
has been recombination (Fig. 1).
Infected mother
Non-infected mother
Non-recombinant
offspring
B
A
A
B
a
b
Recombinant
offspring
b
A
or
a
The females were injected with one of two
bacterial pathogens (Serratia marcescens or
Providencia rettgeri) or given a sham injection.
By examining tens of thousands of the flies’
progeny, the authors found that infected mothers produced a higher fraction of recombinant
offspring than non-infected mothers. This
effect was seen in four fly strains. Infection
with a parasitoid wasp (Leptopilina clavipes)
also induced an increase in recombinant
progeny. Unlike the bacterial experiments, in
which reproductive adults were infected, the
parasitoid wasp infects fruit-fly larvae and
the parasites must be killed for the larva to
survive to adulthood. Thus, in this situation,
the infection is cleared long before meiosis (the
cell division necessary to produce gametes and
during which recombination occurs).
An increase in the observed frequency of
recombinant progeny from infected mothers
could be due to an increase in the recombination rate or to transmission distortion (for
example, if recombinant chromosomes are
more likely than non-recombinant chromosomes to end up in successful gametes). To
tease these possibilities apart, Singh et al. made
use of the fact that exchange of chromosomal
material (crossover events) occurs 4–5 days
before eggs are laid. In their bacterial-infection
experiments, the authors found an increase in
recombinant progeny even in the first 4 days
after the mothers were infected. This rapid
response points to transmission distortion.
A remaining challenge will be to understand
how this distortion occurs. Are recombinant
chromosomes less likely than non-recombinant ones to end up in polar bodies, the small
or
b
a
B
Figure 1 | Frequency of recombinant offspring altered by infection. Diploid organisms, such as fruit
flies and humans, have two copies of each chromosome, which can vary in DNA sequence (represented
by A versus a and B versus b) in every cell except gametes (sperm and egg cells). Gametes contain
only one copy of each chromosome, such that fertilization results in two copies again in the offspring.
The sequence in the offspring can be the same as the parental chromosome, or an exchange of genetic
material between the two chromosomes during gamete production — recombination — can result in
different sequences. Singh et al.1 show that fruit-fly mothers that are infected with parasites produce more
recombinant offspring than uninfected mothers.
4 6 4 | NAT U R E | VO L 5 2 5 | 2 4 S E P T E M B E R 2 0 1 5
© 2015 Macmillan Publishers Limited. All rights reserved
NEWS & VIEWS RESEARCH
cells that are formed during meiosis but do
not transmit genes to future generations? Are
gametes bearing recombinant chromosomes
more viable, or do they somehow outcompete
gametes that have non-recombinant chromosomes? What mechanism might mediate such a
bias? Furthermore, Singh and colleagues’ study
focuses on a single genomic region — it will be
of interest to assess whether similar responses
occur elsewhere in the genome, and if not, why
this region is particularly responsive.
Previous work has demonstrated that pathogens increase recombination in plants during
meiotic and somatic (non-gamete) division7–9.
That the few existing examples of this phenomenon span plants and animals suggests that
pathogen-induced increases in the recombinant fraction could be widespread, although
perhaps achieved through different means,
for example transmission distortion in flies
but higher recombination in plants. If so, does
this intriguing connection between pathogens and natural variation in recombination
constitute convincing evidence to support the
Red Queen hypothesis?
Changes in the proportion of recombinant
offspring in flies and other organisms have
been reported in response to various types of
environmental stress (such as temperature,
nutrition and social stress; reviewed in ref. 10),
although rarely with the rigour of Singh and
colleagues’ work. Is selection by parasites
a driver of the evolution of plasticity in the
recombinant fraction, perhaps one that spills
over to other types of stress? Or is the observed
response to pathogens a by-product of whatever causes plasticity in response to these other
stresses? A crucial first step towards answering
these questions would be to obtain evidence —
so far lacking — that recombinant offspring
are less likely to become infected than nonrecombinant offspring.
Although plasticity in the recombinant
fraction has been known for around 100 years,
it is still poorly studied. We have only the
crudest picture of what conditions alter the
recombinant fraction, by how much and in
which genomic regions. Moreover, theoretical
models10 suggest that the evolution of recombination plasticity is not easily explained for
‘normal’ stresses in diploid organisms (those
that have two copies of each chromosome,
C OMPUTATIO NAL ASTRO PH YSI C S
Monstrous galaxies
unmasked
The enigma of how the most luminous galaxies arise is closer to being solved. New
simulations show that these are long-lived massive galaxies powered by prodigious
gas infall and the recycling of supernova-driven outflows. See Letter p.496
R O M E E L D AV É
T
hree billion years after the Big Bang,
the Universe was a different place from
today. During that epoch, known as
cosmic noon, the average star-formation rate
across the cosmos was 100 times higher than
it is at present, and individual galaxies were
growing commensurately rapidly. This was
illustrated by the surprising discovery1, more
than a decade ago, of galaxies whose starformation rates during that era were
1,000 times the Milky Way’s current output —
no such galaxies are seen in the present-day
Universe. On page 496 of this issue, Narayanan
et al.2 present numerical simulations that offer
unprecedented clarity in understanding the
origins of such deep-space monsters.
These galaxies have extreme properties
and are the most luminous in the Universe.
However, despite their enormous total
energy output, they are faint at optical wavelengths: most of the radiation emitted by their
stars is absorbed by a ‘mask’ of interstellar
dust and re-emitted at longer wavelengths.
Consequently, they remained undiscovered
until the advent of surveys at submilli­metre
and radio wavelengths 1. The very existence of these submillimetre galaxies (SMGs)
presented a challenge to models of galaxy
formation in a cosmological framework, and
has since sparked a vigorous debate in the field
of galaxy-formation theory.
Two schools of thought emerged, centred
around the ‘merger-starburst’ and the ‘smoothaccretion’ hypotheses, respectively3–6. The
former proposes that a given SMG is the
product of a collision between two gas-rich
disk galaxies — this process drives a shortlived (about 108 years) but spectacular burst of
star formation during the galaxies’ coalescence.
The latter argues that SMGs represent the most
massive members of the entire galaxy population, being long-lived phenomena that are continuously fed by gas accretion over periods of
about 109 years.
The merger-starburst hypothesis stems
from a scaled-up analogy of the observation
including flies). Even the seemingly intuitive
Red Queen interpretation of Singh and colleagues’ results is questionable because offspring will always receive half the alleles
carried by their mother, regardless of whether
they are recombinant or not. Although studies
such as this shed light on variation in recombination, there is a long way to go in terms
of fully describing this variation and understanding it from both a mechanistic and an
evolutionary perspective. ■
Aneil F. Agrawal is in the Department of
Ecology & Evolutionary Biology, University of
Toronto, Toronto, Ontario M5S 3B2, Canada.
e-mail: [email protected]
1. Singh, N. D. et al. Science 349, 747–750 (2015).
2. Bell, G. & Maynard Smith, J. Nature 328, 66–68
(1987).
3. Jaenike, J. Evol. Theory 3, 191–194 (1978).
4. Hamilton, W. D. Oikos 35, 282–290 (1980).
5. Peters, A. D. & Lively, C. M. J. Evol. Biol. 20,
1206–1217 (2007).
6. Agrawal, A. F. Evolution 63, 2131–2141 (2009).
7. Kovalchuk, I. et al. Nature 423, 760–762 (2003).
8. Lucht, J. M. et al. Nature Genet. 30, 311–314 (2002).
9. Andronic, L. Can. J. Plant Sci. 92, 1083–1091 (2012).
10.Agrawal, A. F., Hadany, L. & Otto, S. P. Genetics 171,
803–812 (2005).
that the most luminous galaxies in the presentday Universe are almost always involved in
spectacular collisions7. The smooth-accretion
hypothesis is founded on the prediction3 that,
at early cosmic epochs, galaxies were accreting gas at extremely high rates — they could
thus potentially sustain their excessive starformation activity.
Neither scenario has been successful in fully
replicating the observed properties of SMGs.
Researchers have been unable to reproduce
the number of systems required to match
the observations under the merger-starburst
hypothesis6, because collisions between sufficiently large galaxies were rare at those early
times. An influential study4 has argued that
collisions between more-numerous low-mass
galaxies could lead to the formation of SMGs,
but only under the assumption that the starformation process stimulated by the mergers
was heavily weighted towards the production
of massive stars. However, this assumption was
subsequently disfavoured by observational
results8.
Similarly, the smooth-accretion scenario has
been tested in cosmological simulations that
generated SMGs matching those observed, but
they did not reproduce the high luminosities of
these systems5,6. Given that SMGs are thought
to be the progenitors of the well-studied
elliptical galaxies found in the present-day
Universe, the inability to fit these sources
straightforwardly into a cosmological galaxyformation context has been worrying.
In the current study, Narayanan et al. present
a hydrodynamic simulation in a cosmological framework that yields the first SMG
with a luminosity that is a good match to the
2 4 S E P T E M B E R 2 0 1 5 | VO L 5 2 5 | NAT U R E | 4 6 5
© 2015 Macmillan Publishers Limited. All rights reserved